Abstract This project seeks to develop the next generation pro-CARs that are inactive in normal tissues but selectively activated in the tumor microenvironment (TME). The project is based on our previous development of a novel class of chimeric antigen receptors on the basis of single domain antibody mimics that recognize ErbB family members (SDAErbB) on the surface of cancer cells. We propose two specific aims in this project. The first aim is to develop a pro-SDAErbB-CAR that is inactive in normal tissues due to the blockade of the antigen-binding sites by an N-terminal prodomain, but selectively activated in TME through proteolytic cleavage of the prodomain by matrix metalloproteinase MMP-2/-9 highly expressed in TME or by fibroblast activation protein-? (FAP) abundant in the reactive tumor stromal fibroblasts. The second aim is to develop a novel class of CAR that integrates a chemotactic chemokine, which upon being delivered by CAR T-cells and locally released to TME, will attract endogenous cytotoxic T cells (CTLs) to the tumor site to provide higher and enduring antitumor efficacy in a TME-specific manner. The two complementary CAR T-cell platforms have significant advantages on high efficacy and specificity to tumor cells but very low side toxicity and improved safety to normal cells, and therefore the great potential to be applied to treat numerous solid tumors. Although we focus on EGFR- and HER3-specific CARs in this project, the same technology platform can be applied to develop numerous other CARs that target neoantigens, for the treatment of solid tumors of interest.